Paperclip

Paperclip coordinates AI agents for people managing work across development, research, marketing and other business functions. It gives the team a shared structure and keeps its work tied to organization goals.

• Agent teams: Combine Claude Code, Codex, Cursor, OpenClaw and other runtimes in an org chart with roles and reporting lines, or adapt ready-made teams to your work.
• Goals and recurring work: Link tasks and projects to business goals, delegate work between agents, and start recurring tasks through schedules, API calls or webhooks.
• Task history and feedback: Keep conversations, plans, files and run histories attached to tasks, inspect tool calls, and comment on specific passages in documents agents produce.
• Budgets and approvals: Track spending by agent, task, project or goal, pause work at budget limits, approve hires and strategies, and reassign or stop agents.
• Connections and shared skills: Connect GitHub, Notion, Railway or your own MCP server with controls over agent access and action approvals, and create or test skills shared across the team.

Paperclip is MIT-licensed open source. It runs locally or on a self-hosted server without a Paperclip account. Local deployments store application data in local files and an embedded Postgres database; you can also use your own Postgres. One deployment can keep separate organizations' tasks, agents, permissions and histories apart.

You supply the agents and runtimes. Their provider connections determine where model requests go. Budget enforcement uses reported spending, so delayed usage reports and work already in progress can delay a stop. Paperclip's approval controls govern its own workflows; you remain responsible for securing the agents' execution environments.